Food Partitioning of Leaf-Eating Mangrove Crabs (Sesarminae): Experimental and Stable Isotope (13Cand15n) Evidence

Estuarine, Coastal and Shelf Science 87 (2010) 583e590

Contents lists available at ScienceDirect

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier.com/locate/ecss

Food partitioning of leaf-eating mangrove crabs (Sesarminae): Experimental and stable isotope (13Cand15N) evidence

Ditte K. Kristensen a, Erik Kristensen a,*, Perrine Mangion b a Institute of Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark b Vrije Universiteit Brussel, Department of Analytical and Environmental Chemistry, Pleinlaan 2, B-1050 Brussels, Belgium article info abstract

Article history: The feasibility of mangrove leaves as a full diet for sesarmid crabs has been questioned for decades. Since Received 18 October 2009 these leaves are nitrogen-poor, sesarmids probably obtain nitrogen from other sources to sustain Accepted 19 February 2010 growth. The aim of this study was to assess the food partitioning of the sesarmid species Neoepisesarma Available online 1 March 2010 versicolor with emphasis on nitrogen allocation. The preference for animal tissue when crabs were pre- fed diets of different nitrogen content was determined in the laboratory. Furthermore, the possible in Keywords: situ diet composition of N. versicolor was established from carbon and nitrogen stable isotope signature Sesarmid crab (d13C and d15N) of freshly caught individuals and their potential food sources, using a concentration- mangrove fi fi food source dependent mixing model. N. versicolor showed signi cantly higher feeding preferences for sh meat stable isotopes when pre-fed leaf material without than with access to meat, indicating that this crab species can meet mixing model its nitrogen demand by ingesting animal tissue. The stable isotope mixing model based on in situ materials suggests that the diet of N. versicolor consists of w60% leaves in terms of biomass, leaving w40% for other sources such as animal tissue and benthic microorganisms. The biomass contribution from animal tissues, in form of e.g. other crustaceans and fish carcasses, was found to account for w15%. Despite the relative low biomass fraction, animal food sources may contribute with up to half of the nitrogen in the diet of N. versicolor. The quantity of ingested sediment most likely exceeds that of animal tissues. However, due to the low concentration of assimilable microalgae and other microorganism, we propose that sediment associated sources are less important as a nitrogen source for N. versicolor than hitherto presumed. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction et al., 2002). In addition, the crabs facilitate decomposition of mangrove detritus and recycling of nutrients by macerating and Leaf-eating sesarmid crabs are known to have a high impact on fragmenting the litter through ingestion and gut passage. The crab the mangrove ecosystem in the Indo-West Paciﬁc due to their fecal material consists of macerated and incompletely digested leaf distinctive behavior and feeding ecology (for review, see Lee, 1998 litter, which provides easier access for colonization by bacteria and or Kristensen, 2007). Several studies have shown that these crabs other organisms of the detritus food chain (Kristensen and Pilgaard, consume mangrove leaves as their major food source (e.g. 2001). Robertson, 1986; Steinke et al., 1993; Thongtham and Kristensen, Fresh mangrove leaves are unpalatable to most herbivores 2005), and stomach analyses reveal that leaf fragments account because of their high concentrations of refractory and indigestible for 55e95% of the total content (Malley, 1978; Dahdouh-Guebas polyphenolic compounds, such as tannins. Moreover, due to et al., 1999; Thongtham et al., 2008). Sesarmid crabs primarily a substantial content of cellulose and lignin, the C:N ratio is high consume leaves at the sediment ﬂoor or pull them into their compared to plant materials of marine origin. The C:N ratio varies burrows for later ingestion. Hence, they are capable of removing among mangrove species but is in general about 50e100 (e.g. 30e90% of the annual litter production (Robertson, 1986; Micheli, Robertson, 1988; Kristensen et al., 1995). This far exceeds the value 1993b; Slim et al., 1997), reducing tidal exports of mangrove of 17, which is suggested as the maximum for sustainable animal derived organic matter considerably (Slim et al., 1996; Olafsson nutrition (Russell-Hunter, 1970). It has been proposed that the observed preference of sesarmid crabs for partly decomposed leaves is caused by loss of tannins and structural carbon combined * Corresponding author. with enhanced nutritional (i.e. nitrogen) value by microbial colo- E-mail address: [email protected] (E. Kristensen). nization during aging (e.g. Lee, 1989; Robertson and Daniel, 1989;

Micheli, 1993a). However, this hypothesis has been questioned (e.g. The subfamily Sesarminae is very diverse regarding morphology Skov and Hartnoll, 2002; Thongtham and Kristensen, 2005), and and habitats, though most of the members are semi-terrestrial and even partly degraded mangrove leaves are probably insufficient to associated with mangrove ecosystems. This study focused on the maintain crab growth. This is further corroborated by stable isotope species N. versicolor, which plays an important ecological as well as analyses, which show that sesarmid crabs are more enriched in 13C a socio-economical role in mangrove areas of Thailand (Thongtham than mangrove leaves (Bouillon et al., 2002, 2004). The logical et al., 2008). The aim was to investigate the food partitioning of consequence is that leaf-eating crabs most likely supplement their N. versicolor, with emphasis on nitrogen sources. The hypotheses leaf diet with nitrogen-rich food sources to meet the nitrogen tested were: 1. N. versicolor is capable of supplementing its diet demand. But the origin of these alternative nitrogen sources has with animal tissue; and 2. animal tissue contribute substantially as been and still is under debate. a nitrogen source for N. versicolor. These hypotheses were tested in It is a common opinion in the literature that the majority of leaf- laboratory feeding experiments supplemented with stable isotope eating sesarmid crabs augment their nitrogen supply by ingesting composition analyses of wild caught crabs and their possible food sediment and assimilating the associated microalgae and bacteria sources. The most likely elemental composition and contribution of (e.g. Robertson, 1986; Bouillon et al., 2002; Skov and Hartnoll, animal tissue in the diet were determined by applying a concen- 2002). This is based on the observation that many sesarmid crabs tration-dependent stable isotope mixing model. spend a considerable amount of time feeding on sediment (e.g. Micheli, 1993b; Skov and Hartnoll, 2002; Thongtham et al., 2008). 2. Materials and methods In support of this contention, Bouillon et al. (2002) demonstrated that stable carbon isotope ratios of sesarmids and sediment 2.1. Study area correlate. However, Thongtham and Kristensen (2005) argued that the large sesarmid species, Neoepisesarma versicolor, is physically Individuals of N. versicolor and samples of their potential food unable to consume enough sediment to cover its nitrogen demand. sources were collected in the Bangrong mangrove forest (8 030 N, Along this line of evidence, it has been suggested that some ses- 98 250E) located on the north east coast of Phuket Island, Thailand armid crabs primarily achieve their nitrogen from occasional (Fig. 1). Bangrong is a 2.5 km2 fringe forest receiving no river consumption of animal tissue, such as carcasses of fish, dead and discharges and with the only freshwater supply originating from live crustaceans (Thongtham and Kristensen, 2005; Thongtham direct precipitation and run-off from land. The creek area covers et al., 2008), and small invertebrates associated with the sedi- 0.4 km2 and consists primarily of a 3 km longitudinal main tidal ment floor and leaf litter (Kwok and Lee, 1995). The reliance on channel. Tidal range in the area varies from 1 m at neap tide to 3 m animal tissue is supported by findings of crustacean exoskeleton at spring tide. The climate is monsoonal with a wet season from remains and fish scales in the stomach of N. versicolor (Thongtham May to November and a dry season from December to April. The et al., 2008) and Neosarmatium smithii (Giddins et al., 1986). annual precipitation is about 2300 mm and the average

Fig. 1. Map of the Bangrong mangrove forest, Thailand. The two sampling sites are indicated. D.K. Kristensen et al. / Estuarine, Coastal and Shelf Science 87 (2010) 583e590 585 temperature is 28 C. The dominant mangrove species are Rhizo- visible smaller crabs of various species. Green and yellow leaves phora apiculata, Rhizophora mucronata, Ceriops tagal, and Xylocarpus were handpicked from the trees, while brown leaves were granatum. The dominant benthic species are fiddler crabs (familie: collected from the forest floor between trees. Surface sediment Ocypodidae) and sesarmid crabs (family: Grapsidae). (<1 mm) was obtained by careful scraping with a spatula. All samples were transported to the laboratory and immediately 2.2. Neoepisesarma versicolor feeding experiment frozen (18C). Brown leaves were sorted into two groups before freezing; one was gently rinsed in tap water to remove surface The selectivity and satiation of N. versicolor to animal tissue (i.e. debris and the other was not rinsed. Within a few days, the herring meat) when pre-fed food sources of different nitrogen samples were thawed and oven dried for at least 24 h at 80C content was examined in a laboratory experiment carried out at (fauna and leaves) or 105C (sediment). Subsequently the Phuket Marine Biological Center in September and October 2007. samples were grinded to fine powder in an agate mortar. Only The experimental set up consisted of three types of microcosms in muscle tissue from the chelae of crabs was used. The sample which crabs initially had access to: 1) leaves only, 2) leaves and preparation for 13C and elemental (C:N) analysis were completed sediment, and 3) leaves and fish meat. Each microcosm was con- by soaking the subsamples in diluted HCl to remove possible structed by a glass aquarium (29 29 33 cm) angled so that 2/3 of carbonates and redried. As this treatment is reported to affect the bottom was covered with seawater to a maximum depth of 15N-values (Bunn et al., 1995), subsamples for 15N analysis were 2 cm. Full strength seawater was diluted with freshwater to not acidified. a salinity of about 20 as a precaution to prevent high salinities due Organic carbon and nitrogen content as well as stable isotope to evaporation. No tidal rhythm was simulated to avoid disturbing composition (12C, 13C, 14N and 15N) of fauna, leaf and sediment the crabs. The water was changed every five days to maintain an samples were determined by combusting pre-weighed subsamples acceptable water quality. Mesocosms were maintained at air in a Thermo Scientific Flash1112 elemental analyzer coupled to temperature (28e31 C) and exposed to subdued sunlight. The a Thermo Scientific Delta V via a conflo III interface. Stable isotope humidity was constantly around 98%. A total of nine crabs were analyses of carbon and nitrogen were analyzed separately to avoid used in the three types of microcosms, providing three replicates of acidification errors and to obtain more accurate measurements for each. All crabs were intact non-moulting adult males with a cara- sediment and mangrove leaves, which usually exhibit high C:N pace width of 3e4 cm. ratios. Stable isotopes ratios are expressed as d values (&) relative Crabs were kept on the different pre-diets for extended time to the conventional standards (VPDB limestone for C and atmo- periods prior to the final meat-feeding observations. All crabs spheric N2 for N) according to: were offered fresh leaves of R. apiculata, which is a dominant R R mangrove species at their habitat. New leaves were given and dX ¼ sample standard 103½& uneaten remains were removed every second day during the pre- Rstandard feeding period. Crabs in type 1 and type 2 mesocosms were pre- where X represents 13Cor15N and R is 13C/12C in the case of carbon fed leaves for 17 days, while this treatment in the type 3 meso- and 15N/14N in the case of nitrogen. cosm only lasted 5 days. The dry part of the bottom in the type 2 mesocosm was covered initially with 3 cm deep layer of freshly collected mangrove surface sediment. The sediment was not 2.4. Statistical analysis renewed during the entire experiment. Crabs in the type 3 mesocosm were offered fish meat (w20 g herring) 4 days prior to the Means within the feeding experiment were compared by two- final observations. All remains of fish meat were removed one day sample t-tests. However, the group, which had a pre-diet of leaves fi later, leaving these crabs with leaves as the only food source for 3 and sh, did not have any variation due to a response variable of days. zero in all replicates. Consequently, the comparisons with this The response variable of the final observations was crab feeding group were done by one-sample t-tests for means against zero. The fi preference for animal tissue. Each crab was offered a piece of a fresh signi cance level was set at 5%. An a priori Sequential Bonferroni herring at dusk, which is within their natural feeding period, and correction was made on the succeeding t-tests to avoid the risk of the total time spent feeding was observed for 40 min. The regis- type II errors. tration time did not include the first 10 min habituation time of the crabs after transferring the fish. Feeding was defined as the time 3. Results where crabs had physical contact with the fish. Observations were carried out by two quiescent persons wearing head torches. The 3.1. Feeding experiment light beam did not seem to disturb the crabs. The response variable was expressed as percent eating time during the final 30 min Despite the simple design and low sample size (n ¼ 3) of the registration period. feeding experiment, the results are very informative. All specimens of N. versicolor behaved well with continuous ingestion of 2.3. In situ stable isotope signatures leaves during the experiment and survival was 100%. The two groups of crabs with a pre-diet of leaves only, and leaves and Carbon and nitrogen stable isotope signatures of N. versicolor sediment spent most of the final registration period handling and and its possible food sources were used to identify the parti- ingesting fish meat (Table 1). There was no significant difference tioning of these food items in its diet. Fauna, vegetation, and in fish-eating time between these two groups, but only the group sediment samples for stable isotope analysis were collected at on sole leaf pre-diet was significantly different from zero. Simi- two sites along the main creek in the Bangrong mangrove forest larly, the individuals that were offered fish meat for 24 h three in December 2008. Site RM was located within a monostand of R. days prior to the final observations all consumed this food source mucronata close to an open creek bank, while site RA was willingly (on average w0.5 g). However, none of these touched the selected about 200 m away within a monostand of R. apiculata fish they were offered during the final 30 min registration period. that covered the creek bank completely (Fig. 1). Six individuals of Access to sediment in the microcosms did not affect the crabs’ N. versicolor were caught by hand at each site together with all willingness to feed on fish meat. 586 D.K. Kristensen et al. / Estuarine, Coastal and Shelf Science 87 (2010) 583e590

Table 1 Percentage total eating time of Neoepisesarma versicolor when offered a piece of fresh fish for 30 min (mean standard deviation, n ¼ 3). Crabs were kept on different diets for various time periods prior to the observations; 17 days for leaves only or leaves and sediment and 5 days for leaves and fish. Values with the same letter are not significantly different (p > 0.05).

Pre-diet Time on diet (days) Total feeding time (%) Leaves 17 92 11 a Leaves and sediment 17 74 39 ab Leaves and ﬁsh 5 0 0b

3.2. Stable isotope signatures

Leaves of the two Rhizophora species from the two sites showed comparable average d13C values irrespective of degree of senes- cence, decay and rinsing (Fig. 2). However, the favorite food of N. versicolor, partly degraded brown leaves covered with debris, was about 2 & more depleted in 15N than green leaves of both R. mucronata and R. apiculata. Surface sediment was slightly enriched 13 15 & in both C and N compared with mangrove leaves (about 26 Fig. 2. d13C and d15N signatures of mangrove leaves at various stages of aging and for C and 5 & for N; Fig. 3). Unfortunately, all our efforts in decomposition. Rm-g and Ra-g indicates green leaves of Rhizophora mucronata and R. extracting benthic microalgae without sediment contamination apiculata, respectively. The letters y, b and b þ represent yellow leaves, brown leaves failed. and brown leaves with debris, respectively. The broken lines trace the change in isotope signature during aging and decay for each species. Error bars indicate standard The ocypodid crabs Uca paradussumieri and Uca forcipata, and deviation (n ¼ 4e6). the varunid crab Metaplax elegans were the most abundant potential prey items for N. versicolor at the two sampling sites. They were, however, much more common at the open site RM than the Both groups, with pre-diets of leaves only and leaves and sedi- densely vegetated site RA. The few individuals caught at site RA ment, respectively, were undoubtedly hungry for meat after 17 days were unfortunately too small to obtain sufﬁcient muscle samples on nitrogen-deprived diets. Thus, the microorganisms of the for analysis. The stable isotope signature of these three species from mangrove sediment did in this case not contribute signiﬁcantly to site RM were similar with d13C ranging from 16.6 to 18.0 & and the crabs nitrogen demand. Although the results indicate that d15N ranging from 7.5 to 8.9 & (Fig. 3). Due to the lack of specimens N. versicolor does not rely on bacteria as a main nitrogen source, the from site RA, we assume similar signatures here in the subsequent role of benthic microalgae as a food source cannot be interpreted data analysis. from this experiment because it was maintained under dim light The d13C signature of N. versicolor was similar at site RM and RA conditions. (24.3 &), which is roughly 5& enrichment compared to Rhizo- Animal tissue is not expected to be a frequent food source for phora leaves (Fig. 3). Conversely, d15NofN. versicolor was 3e4& sesarmid crabs in the wild, and they are presumably able to cope higher than mangrove leaves and notably 1& higher at site RM than site RA (7.6 and 6.7&, respectively), suggesting a trophic difference between the two sites. 12

4. Discussion

4.1. Food preference and nitrogen source of N. versicolor Uf Me 8 Nv-RM

Fish meat is potentially a nitrogen-rich food source for the ) Up crabs and the fact that they willingly ingested it in the feeding oo /

o Nv-RA

experiment supports our hypothesis that animal tissue can be ( a supplementary food source for N. versicolor. The difference in N feeding behavior among the three experimental groups clearly 15 Rm Sed demonstrates that the preference of crabs for animal tissue 4 depends on the pre-diet. Crabs can apparently be satiated with Ra regard to nitrogen after a meal of meat and our experiment showed that such satiation lasts for at least three days. BMA A reasonable explanation for this behavior is that N. versicolor can store excess nitrogen intracellularly, as suggested by Thongtham 0 -35 -30 -25 -20 -15 -10 and Kristensen (2005). A similar mechanism is known from 13 several species of land crabs, which store nitrogen intracellularly C (o/oo) as uric acid in spongy connective tissue cells throughout the body (Gifford, 1968; Wolcott and O’Connor, 1992; Linton and Fig. 3. d13C and d15N signatures of brown leaves with debris (Rm e Rhizophora Greenaway, 1997). The idea of nitrogen storage is consistent mucronata and Ra e R. apiculata, n ¼ 5), sediment (Sed, n ¼ 7)), benthic microalgae (BMA, n ¼ 7(15N) or 15 (13C), potential crab prey for Neoepisesarma versicolor (Up e with our observations of sesarmid crabs in the wild, where their Uca paradussumieri, n ¼ 10; Uf e Uca forcipata, n ¼ 4; and Me e Metaplax elegans, tendency to act as predators is unpredictable (personal n ¼ 3) and N. versicolor (Nv-RM e individuals from site RM and Nv-RA e individuals observation). from site RA, N ¼ 6). Error bars indicate standard deviation. D.K. Kristensen et al. / Estuarine, Coastal and Shelf Science 87 (2010) 583e590 587 with long periods without ingestion of meat. This is supported by Table 2 the feeding experiment, since there were no visible signs that Stable carbon and nitrogen isotope signatures of benthic microalgae (BMA) from various mangrove locations. The global average (standard deviation) used in the the crabs were suffering after 17 days on a restricted diet. On the mixing model is shown. other hand, we did not assess changes in crab biomass during the d13 & d15 & experiment and it cannot be ruled out that the animals lost Location C( ) N( ) Reference weight. However, a long term study on two sesarmid species Molokai, Hawaii 20.5 Demopoulos et al. (2007) (Chiromanthes bidens and Parasesarma plicata) showed that these Jobos Bay, Puerto Rico 21.5 Demopoulos et al. (2007) Laguna Joyuda, Puerto Rico 19.9 2.2 France (1998) crabs are actually able to grow, but not reproduce, on a diet of Estero Morales, Costa Rica 21.0 3.9 Dittel et al. (1997) pure mangrove leaves (Kwok and Lee, 1995). Further experi- Gazi, Kenya 24.5 1.4 P. Mangion, unpubl mental studies are needed to conﬁrm the necessity of animal Ras Dege, Tanzania 21.2 1.3 P. Mangion, unpubl tissue as a food source for sesarmids and their capacity to store Zanzibar 12.4 M. skov, unpubl Coringa Sanctuary, India 17.3 1.7 Bouillon et al. (2002) excess nitrogen. In any case, it appears that N. versicolor obtains Coringa Sanctuary, India 18.5 1.7 Bouillon et al. (2004) much of its nitrogen requirement from animal sources. This Pearl River Estuary, Hong 19.3 Lee (2000) species may to some degree also supplement its nitrogen intake Kong with benthic microalgae or small invertebrates associated with Sikao Creek, Thailand 17.9 1.2 Kon et al. (2007) the sediment. This could explain why N. versicolor spends Tanshui Estuary, Taiwan 19.4 Hsieh et al. (2002) Tanshui Estuary, Taiwan 19.3 Hsieh et al. (2002) considerable time foraging on sediment (Thongtham unpub- Moreton Bay, Australia 24.4 Guest et al. (2004) lished). Taken the diversity of feeding habitats and habits of Moreton Bay, Australia 20.4 3.8 Connolly (2003) various sesarmid crabs (Lee, 1998) into consideration it seems Moreton Bay, Australia 23.7 Guest and Connoly feasible that they may have different strategies for obtaining (2004) Average 20.1 3.0 2.2 1.1 nitrogen.

4.2. Stable isotope signatures of N. versicolor and its potential food 4.3. Dietary composition of N. versicolor sources Stomach content analysis (Thongtham et al., 2008) and stable The stable carbon isotope signatures found for leaves (mean isotope surveys (this study, Bouillon et al., 2002) both suggest that N. 30.1 &, Fig. 2) are within the range reported for mangroves versicolor has a heterogeneous and mixed diet composed of leaves around the world, 35 to 22 & (Bouillon et al., 2008), mangrove leaves, animal tissue in the form of invertebrates and fish and are consistent with other measurements on the genus Rhi- carrions, and benthic microorganisms associated with sediment. zophora (Bouillon et al., 2004). The differences in d15N among the The relative importance of these food sources can be assessed using various leaf categories suggest that the leaf 15N pool is diluted stable isotope signatures and mixing models. In general, the during leaching, aging and decomposition (Fry and Smith, 2002; proportion of n þ 1 different food sources can be found uniquely by n Werry and Lee, 2005). Bacteria colonizing the leaves during different elemental isotopic signatures with linear mixing models decomposition may thus assimilate 15N depleted inorganic based on Euclidean-distances (e.g. Ben-David et al., 1997) or mass nitrogen from the surroundings or through nitrogen fixation (Gu, balance equations (e.g. Phillips and Koch, 2001). Other models that 2009). cope with unlimited number of sources have been developed as Stable isotope signatures of N. versicolor from Bangrong are well. Among these, the IsoSource model (Phillips and Gregg, 2003) consistent with measurements of the sister sesarmid species, determines boundaries for the contributions of each source by Episesarma versicolor, from India (Bouillon et al., 2002, 2004). The examining all combinations of food sources that sum to the observed roughly 5& enriched d13C signature of N. versicolor compared to mixture of isotopic signatures. MixSir (More and Semmens, 2008; Rhizophora leaves, confirms that this sesarmid species does not feed but see Jackson et al., 2009) and SIAR (Panell et al., 2008) are new solely on mangrove litter (Fig. 3). Bayesian-mixing models based on IsoSource, but with the great Although surface sediment was slightly enriched in both 13C and advantage of incorporating uncertainty and prior information. 15N compared with mangrove leaves, it is not realistic to consider It is important, however, that the general assumptions of mixing bulk sediment as a food source for N. versicolor. It is known that models do not violate the biological reality when evaluating the other deposit-feeders ingesting bulk sediment primarily assimilate food sources of an omnivore such as N. versicolor. Especially the benthic microalgae (BMA) and to some degree bacteria, while the model assumption of equal proportions of C and N assimilated from detrital fraction is largely undigestable (Andresen and Kristensen, each food source is critical when dealing with omnivores (also 2002). Since our effort in extracting BMA without sediment noted by Stenroth et al., 2006). As pointed out by Gannes et al. contamination failed, we instead gathered a global average signa- (1997) meat is likely the main source of proteins and lipids for an ture of various benthic microalgae from the literature for use in our omnivore, whereas plant material is the main source of carbohy- modeling considerations (Table 2; Fig. 3). drates. The same pattern is expected for N. versicolor as mangrove The observed d13C values of the potential prey, U. para- leaves are extremely poor in nitrogen, having C:N ratios around 30 dussumieri, U. forcipata, and M. elegans, are comparable to those and 15 times higher than potential animal sources and microalgae, found for the same genera in Kenya and India, while d15N is higher respectively. Hence, bulk nitrogen is expected to originate from than in Kenya and similar to those from India (Bouillon et al., 2004). digested animal tissue and to some extend microorganisms. Both Uca species and M. elegans are sediment surface grazers with Only one mixing model deals with unequal assimilation of C and BMA as their primary food source (France, 1998; Weis and Weis, N, namely the concentration-dependent IsoConc model developed 2004; Kon et al., 2007). However, 13C and 15N of these crabs are by Phillips and Koch (2001). The model assumes that “for each enriched 2e3& and 5e6&, respectively, compared with BMA, element, a source’s contribution is proportional to the contributed which exceeds the global mean fractionation of 0.4& and 2.3& per mass times the elemental concentration in that source”. This model trophic level for a large range of aquatic animals (McCutchan et al., is recommended whenever the elemental composition varies 2003). These crabs may therefore also consume BMA grazing substantially among food sources. However, the model has been micro- and meiofauna as suggested for Uca spp. by Reinsel (2004). criticized for incorrectly using C and N concentrations of the food 588 D.K. Kristensen et al. / Estuarine, Coastal and Shelf Science 87 (2010) 583e590 itself rather than the fraction assimilated (Koch and Phillips, 2002; The IsoConc model showed that brown leaves, crab tissue, and Robbins et al., 2002). Bearing this limitation in mind, we have BMA provide a realistic combination of food sources for N. versicolor chosen to use an adjusted version of IsoConc model to estimate the (Fig. 4). Brown leaves are the main food in terms of biomass and dietary composition of N. versicolor at the two mangrove habitats. carbon with a contribution of slightly less than 60% at both site RM Fractionation corrections of carbon and nitrogen were set at 0.4& and RA (Table 4). This proportion may, however, be overestimated and 2.3& per trophic level, respectively, as these are global means considering the uncorrected elemental concentrations of leaves. for a large range of aquatic animals (McCutchan et al., 2003). Nevertheless, the result agrees very well with Thongtham et al. Our choice of food source data (Table 3) was based on the (2008), who found that the stomach content of N. versicolor con- knowledge of feeding behavior by N. versicolor. The leaf diet was tained 62% higher plant material, while the remaining 38% were represented by brown leaves only, as N. versicolor (Thongtham composed of detritus and mineral particles. These two independent et al., 2008) and other sesarmids (e.g. Giddins et al., 1986; Lee, observations certify the importance of leaves in the diet of N. ver- 1989) generally prefer leaves at this stage. Green and yellow sicolor. With respect to nitrogen, the situation is more complicated. leaves are inferior due to lower nutritional value and high levels of The proportion of microalgae and animal tissues in the diet is inhibitory compounds (e.g. tannins, Hernes et al., 2001; Kristensen dependent on the habitat of N. versicolor. Our results indicate that et al., 2008). In addition, we selected brown leaves with intact N. versicolor is capable of feeding on the crab fauna present, but also debris layer to imitate leaves as they are ingested by the crabs. that it can switch between nitrogen-rich food sources depending Sediment as a whole was not included as a food source, since on their availability. At site RM, where the prey crab species are inorganic particles and refractory detritus are not assimilated by abundant, microalgae and animal tissues account for 27% and 16%, the crabs. Instead benthic microalgae (BMA) were included as they respectively, of the assimilated biomass compared with 32% and are expected to make up the bulk of the microorganisms assimi- 10% at site RA. As a consequence, the nitrogen budgets differ lated from the sediment. Their stable isotope signature was derived between the two sampling sites. N. versicolor obtains 50% nitrogen from the above mentioned literature survey (Table 2). Lastly, animal from animal tissue at site RM, while the proportion only is 35% at tissue was included as a food source. For this purpose we used the three abundant crab species mentioned above (i.e. U. paradussumieri, U. forcipata and M. elegans). As the model only handles 12 three food sources, we used the average of the three species isotope signatures and C:N ratios. 50% 75% 100% crab The elemental C and N concentrations applied in the IsoConc 25% 10 model were, if possible, corrected for digestibility. C and N in the analyzed animal tissue are assumed to be 100% digestible (Koch and Phillips, 2002), which is a robust assumption as only muscle tissues 8 was used. In contrast, the mass of consumed BMA is not fully digested as these mainly consist of diatoms with indigestible silica frustules (Thongtham et al., 2008), which may account for up to 6 100% leaf 50% of diatom biomass (Canfield et al., 2005). Accordingly, we assumed that the digestible organic tissue of diatoms contains 45% C, as commonly found for microalgae without thick inorganic 4 100% BMA envelopes (e.g. Kristensen, 1990). The corresponding N content was fi Site RM

estimated to 6.8% from a Red eld C:N weight ratio of 6.6:1. )

Digestibility of vascular plant tissues varies among herbivores. oo / 2 Thongtham and Kristensen (2005) reported assimilation efﬁcien- o ( cies for N. versicolor ingesting brown leaves of 6.5% for the dry 12 N matter and around 40% for C and N. Although these results are 15 100% crab inconsistent, they indicate a high assimilation of C and N in 50% 75% 25% proportions similar to the bulk C:N ratio of the ingested leaf 10 material. However, since the assimilated concentrations are not known, we were forced to use the uncorrected C and N concentrations of brown leaves in the model. 8

Table 3 Input data used in the stable isotope mixing model. The* indicates that the values are 6 100% leaf estimated (see text for details).

n d13C d15N [C] (%) [N] (%) C:N 4 100% BMA Study animal N.versicolor (RA.) 6 24.3 6.7 40.8 13.0 3.1 Site RA N.versicolor (RM) 6 24.2 7.6 40.3 12.8 3.1 2 Food sources Brown leaves (RA) 5 28.5 3.3 44.1 0.5 91.6 -30 -25 -20 -15 Brown leaves (RM.) 5 29.3 3.7 38.6 0.4 96.3 13 (o/oo) Benthic microalgae 16/8 20.1 2.2 45.0* 6.8* 6.6 C

U. paradussumieri 10 16.6 7.5 40.4 12.5 3.2 Fig. 4. Concentration-dependent mixing triangles for Neoepisesarma versicolor from U. forcipata 4 18 8.1 39.4 12.5 3.2 sampling site RM (upper) and RA (lower). Isotope signatures for pure diets at the M. elegans 3 16.9 8.9 39.6 12.6 3.1 vertices of the triangles have been corrected for trophic fractionation (0.4& for C and 2.3& for N). The ﬁlled symbols indicate the isotope signature of the consumer Average for crab prey 17 17.2 8.2 39.8 12.5 3.2 (N. versicolor). D.K. Kristensen et al. / Estuarine, Coastal and Shelf Science 87 (2010) 583e590 589

Table 4 Secondly, key information is missing regarding the assimilation Food source partitioning at the two sampling sites RM and RA as estimated by the efﬁciency of different food sources. Thirdly, we used global average concentration-dependent IsoConc mixing model. f indicates fractional contribution fractionation values as no information is available for mangrove of biomass, carbon and nitrogen. crustaceans. Fourthly, we did not account for internal isotopic N. versicolor (RM) N. versicolor (RA) routing (for review, see Gannes et al., 1997), where stable isotopes fleaves are routed differently to various tissues and body compartments. Biomass 0.57 0.58 Finally, stochasticity of stable isotope ratios of consumer and food Carbon 0.54 0.58 Nitrogen 0.06 0.08 sources were eliminated, as the model is deterministic and only provides one solution. fBMA Biomass 0.27 0.32 Carbon 0.30 0.33 Acknowledgements Nitrogen 0.44 0.57

fanimal We thank N. Thongtham and the staff at Phuket Marine Bio- Biomass 0.16 0.10 logical Center, Thailand, for hospitality and logistical support Carbon 0.16 0.09 fi Nitrogen 0.50 0.35 during eld samplings. This study was supported by grant # 09- 071369 from the Danish Research Council. site RA (Table 4). Most of the remaining nitrogen is derived from References BMA (44% at site RM and 57% at site RA) with only limited contribution from leaf material. It appears that the lack of animal prey at Andresen, M., Kristensen, E., 2002. The importance of bacteria and microalgae in the diet of the deposit-feeding polychaete Arenicola marina. Ophelia 56, site RA forces N. versicolor to feed more on organisms from a lower 179e196. trophic level (e.g. BMA) than at site RM. The modeling results Ben-David, M., Flynn, R.W., Schell, D.M., 1997. Annual and seasonal changes in diet support our hypothesis that animal tissues contribute substantially of martens: evidence from stable isotope analysis. Oecologia 111, 280e291. as a nitrogen source for N. versicolor, although the consumed mass Bouillon, S., Koedam, N., Raman, A.V., Dehairs, F., 2002. Primary producers sustaining macroinvertebrate communities in intertidal mangrove forests. is relative small. The food partitioning obtained by the mixing Oecologia 130, 441e448. model agrees well with stomach analyses of N. versicolor. Bouillon, S., Moens, T., Overmeer, I., Koedam, N., Dehairs, F., 2004. Resource utili- Thongtham et al. (2008) found that 11% of the examined crabs zation patterns of epifauna from mangrove forests with contrasting inputs of fi local versus imported organic matter. Marine Ecology Progress Series 278, contained crustacean remains while 7% had sh scales in their 77e88. stomach. Furthermore, benthic diatoms and filamentous algae Bouillon, S., Connolly, R.M., Lee, S.Y., 2008. Organic matter exchange and cycling in occurred in 86% and 47% of the crabs, respectively. It can therefore mangrove ecosystems: recent insights from stable isotope studies. Journal of Sea Research 59, 44e58. be concluded that animal tissue is not likely an every-day-meal for Bunn, S.E., Loneragan, N.R., Kempster, M.A., 1995. Effects of acid washing samples on N. versicolor, but even so it functions as an essential supplementary stable isotope ratios of C and N in penaeid shrimps and seagrass: implications nitrogen source. for food web studies using stable isotopes. Limnology and Oceanography 40, 622e625. The estimated nitrogen contribution of microalgae was higher Canfield, D.E., Kristensen, E., Thamdrup, B., 2005. Aquatic Geomicrobiology. Elsevier than expected. In order to obtain sufficient amounts of algae, Academic Press, Amsterdam, p. 636. N. versicolor must non-selectively ingest a large amount of sedi- Connolly, R.M., 2003. Differences in trophodynamics of commercially important fish between artificial waterways and natural coastal wetlands. Estuarine and ment because it has chelae designed for grasping and not sorting Coastal Shelf Science 58, 929e936. (personal observation). Many specialized mangrove crabs (e.g. Dahdouh-Guebas, F., Giugglioli, M., Oluoch, A., Vannini, M., Cannicci, S., 1999. fiddler crabs) are adapted for the latter method. It has been esti- Feeding habits of non-ocypodid crabs from two mangrove forests in Kenya. e mated that N. versicolor must ingest 7-23 cm3 sediment Bulletin of Marine Science 64, 291 297. 1 1 Demopoulos, A.W.J., Fry, B., Smith, C.R., 2007. Food web structure in exotic and (g ww) crab d to fully cover its nitrogen demand from BMA and native mangroves: a Hawaii-Puerto Rico comparison. Oecologia 153, 675e686. other sediment-based nitrogen-rich items (Thongtham and Dittel, A.I., Epifanio, C.E., Cifuentes, L.A., Kirchman, D.L., 1997. Carbon and nitrogen sources for shrimp postlarvae fed natural diets from a tropical mangrove Kristensen, 2005). Although this species, according to the IsoConc e e system. Estuarine and Coastal Shelf Science 45, 629 637. model, only obtains about 40 50% of the nitrogen from BMA, the France, R., 1998. Estimating the assimilation of mangrove detritus by fiddler crabs in volume of sediment still seems unrealistically high compared to the Laguna Joyuda, Puerto Rico, using dual stable isotopes. Journal of Tropical size of the crab (w25 g ww). If N. versicolor is as inefficient in Ecology 14, 413e425. Fry, B., Smith, T.J., 2002. Stable isotope studies of red mangroves and filter feeders exploiting BMA from the sediment as we assume, it probably from the Shark river Estuary, Florida. Bulletin of Marine Science 70, 871e890. obtains more nitrogen from animal tissues or other easy accessible Gannes, L.Z., O’Brien, M.C., Marínez del Rio, C., 1997. Stable isotopes in animal sources than predicted by the model. The model results may be ecology: assumptions, caveats, and a call for more laboratory experiments. Ecology 78, 1271e1276. misleading due to the lack of stable isotope signatures and Giddins, R.L., Lucas, J.S., Neilson, M.J., Richards, G.N., 1986. Feeding ecology of the elemental concentrations of microalgae from Bangrong. Further- mangrove crab Neosarmatium smithii (Crustacea: Decapoda: Sesarmidae). more, N. versicolor may rely on other food sources than those Marine Ecology Progress Series 33, 147e155. Gifford, C.A., 1968. Accumulation of uric acid in the land crab, Cardisoma guanhumi. included in the model, such as macroalgae, bacteria or small American Zoologist 8, 521e528. invertebrates (e.g. meiofauna). Gu, B., 2009. Variations and controls of nitrogen stable isotopes in particulate In conclusion, our results verify the hypotheses that N. versicolor organic matter of lakes. Oecologia 160, 421e431. can feed on animal tissue and by doing so supplement its dietary Guest, M.A., Connolly, R.M., 2004. Fine-scale movement and assimilation of carbon in saltmarsh and mangrove habitat by resident animals. Aquatic Ecology 38, need for nitrogen. While the feeding experiment illustrates that N. 599e609. versicolor adjusts the intake of animal tissue according to its sati- Guest, M.A., Connolly, R.M., Loneragan, N.R., 2004. Carbon movement and assimi- ation level, the applied mixing model suggests that animal-based lation by invertebrates in estuarine habitats at a scale of metres. Marine Ecology Progress Series 278, 27e34. food sources may cover up to 50% of the nitrogen demand of this Hernes, P.J., Benner, R., Cowie, G.L., Goñi, M.A., Bergamaschi, B.A., Hedges, J.I., 2001. mangrove sesarmid crab when prey is abundant. However, the Tannin diagenesis in mangrove leaves from a tropical estuary: a novel molec- modeling results should be considered as rough estimates because ular approach. Geochimica et Cosmochimica Acta 65, 3109e3122. Hsieh, H.L., Chen, C.P., Chen, Y.G., Yang, H.H., 2002. Diversity of benthic organic of several potential errors. Firstly, BMA isotope signatures and matter flows through polychaetes and crabs in a mangrove estuary: delta C-13 elemental composition are not available from our location. and delta S-34 signals. Marine Ecology Progress Series 227, 145e155. 590 D.K. Kristensen et al. / Estuarine, Coastal and Shelf Science 87 (2010) 583e590

Jackson, A.L., Inger, R., Bearhop, S., Parnell, A., 2009. Erroneous behaviour of MixSIR, Panell, A., Inger, R., Bearshop, S., Jackson, A.L., 2008. SIAR: stable isotope analysis in a recently published Bayesian isotope mixing model: a discussion of More & R. http://cran.r-project.org/web/pakages/siar/index.html. Semmens (2008). Ecology Letters 12, E1eE5. Phillips, D.L., Gregg, J.W., 2003. Source partitioning using stable isotopes: coping Koch, L.P., Phillips, D.L., 2002. Incorporating concentration dependence in stable with too many sources. Oecologia 136, 261e269. isotope mixing models: a reply to Robbins, Hilderbrand, and Farley. Oecologia Phillips, D.L., Koch, P.L., 2001. Incorporating concentration dependence in stable 133, 13e18. isotope mixing models. Oecologia 130, 114e125. Kon, K., Kurokura, H., Hayashizaki, K., 2007. Role of microhabitats in food webs of Reinsel, K.A., 2004. Impact of fiddler crab foraging and tidal inundation on an benthic communities in a mangrove forest. Marine Ecology Progress Series 340, intertidal sandflat: season-dependent effects in one tidal cycle. Journal of 55e62. Experimental Marine Biology and Ecology 313, 1e17. Kristensen, E., 1990. Characterization of biogenic organic matter by stepwise ther- Robbins, C.T., Hilderbrand, G.V., Farley, S.D., 2002. Incorporating concentration mogravimetry (STG). Biogeochemistry 9, 135e159. dependence in stable isotope mixing models: a response to Phillips and Koch Kristensen, E., 2007. Mangrove crabs as ecosystem engineers; with emphasis on (2002). Oecologia 133, 10e13. sediment processes. Journal of Sea Research 59, 30e43. Robertson, A.L., 1986. Leaf-burying crabs: their influence on energy flow and export Kristensen, E., Bouillon, S., Dittmar, T., Marchand, C., 2008. Organic carbon dynamics from mixed mangrove forests (Rhizophora spp.) in northeastern Australia. in mangrove ecosystems. Aquatic Botany 89, 201e219. Journal of Experimental Marine Biology and Ecology 102, 237e248. Kristensen, E., Holmer, M., Banta, G.T., Jensen, M.H., Hansen, K., 1995. Carbon, Robertson, A.L., 1988. Decomposition of mangrove leaf litter in tropical Australia. nitrogen and sulfur cycling in sediment of the Ao Nam Bor Mangrove forest, Journal of Experimental Marine Biology and Ecology 116, 235e247. Phuket, Thailand: a review. Phuket Marine Biological Center Research Bulletin Robertson, A.L., Daniel, P.A., 1989. The influence of crabs on litter processing in high 60, 37e64. intertidal mangrove forests in tropical Australia. Oecologia 78, 191e198. Kristensen, E., Pilgaard, R., 2001. The role of fecal pellet deposition by leaf-eating Russell-Hunter, W.D., 1970. Aquatic Productivity: An Introduction to Some Basic sesarmid crabs on mineralization processes in a mangrove sediment (Phuket, Aspects of Biological Oceanography and Limnology. MacMillan, New York, Thailand). In: Aller, J.Y., Woodin, S.A., Aller, R.C. (Eds.), Organism-Sediment 306 pp. Interactions. University of South Carolina Press, Columbia, pp. 369e384. Skov, M.W., Hartnoll, G.R., 2002. Paradoxical selective feeding on a low-nutrien diet: Kwok, P.W., Lee, S.Y., 1995. The growth performances of two mangrove crabs, Chi- why mangrove crabs eat leaves? Oecologia 131, 1e7. romanthes bidens and Parasesarma plicata under different leaf litter diets. Slim, F.J., Gwada, P.M., Kodjo, M., Hemminga, M.A., 1996. Biomass and litterfall of Hydrobiologia 295, 141e148. Ceriops tagal and Rhizophora mucronata in the mangrove forest of Gazi Bay, Lee, S.Y., 1989. The importance of sesarminae crabs Chiromanthes spp. and inun- Kenya. Marine and Freshwater Research 47, 999e1007. dation frequency on mangrove (Kandelia candel (L.) Druce) leaf litter turnover in Slim, F.J., Hemminga, M.A., Ochieng, C., Jannink, N.T., Cocheret de la a Hong Kong tidal shrimp pond. Journal of Experimental Marine Biology and Moriniere, E., Van der Velde, G., 1997. Leaf litter removal by the snail Ter- Ecology 131, 23e43. ebralia palustris (Linnaeus) and sesarmid crabs in an east African mangrove Lee, S.Y., 1998. Ecological role of grapsid crabs in mangrove ecosystems: a review. forest (Gazi Bay, Kenya). Journal of Experimental Marine Biology and Ecology Marine and Freshwater Research 49, 335e343. 215, 35e48. Lee, S.Y., 2000. Carbon dynamics of Deep Bay, eastern Pearl River estuary, China. II: Steinke, T.D., Rajh, A., Holland, A.J., 1993. The feeding behavior of the red mangrove Trophic relationship based on carbon- and nitrogen-stable isotopes. Marine crab Sesarma meinerti De Man, 1887 (Crustacea: Decapoda: Grapsidae) and its Ecology Progress Series 205, 1e10. effect on the degradation of mangrove leaf litter. South Africa Journal of Marine Linton, S.M., Greenaway, P., 1997. Urate deposits in the Gecarcinid land crab Science 13, 151e160. Gecarcoidea natalis are synthesized de novo from excess dietary nitrogen. Stenroth, P., Holmqvist, N., Nyström, P., Berglund, O., Larsson, P., Granéli, W., 2006. Journal of Experimental Biology 200, 2347e2354. Stable isotopes as indicator of diet in omnivorous crayfish (Pacifastacus Malley, D.F., 1978. Degradation of mangrove leaf litter by the tropical Sesarmid crab leniusculus): the influence of tissue, sample treatment, and season. Canadian Chiromanthes onychophorum. Marine Biology 49, 377e386. Journal of Fisheries and Aquatic Science 63, 821e831. McCutchan, J.H., Lewis, W.M., Kendall, C., McGrath, C.C., 2003. Variation in trophic Thongtham, N., Kristensen, E., 2005. Carbon and nitrogen balance of leaf-eating shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos 102, sesarmid crabs (Neoepisesarma versicolor) offered different food sources. Estu- 378e390. arine, Coastal and Shelf Science 65, 213e222. Micheli, F., 1993a. Effect of mangrove litter species and availability on survival Thongtham, N., Kristensen, E., Purangprasan, S.Y., 2008. Leaf removal by sesarmid moulting, and reproduction of the mangrove crab Sesarma messa. Journal of crabs in Bangrong Mangrove forest, Phuket, Thailand: with emphasis on the Experimental Marine Biology and Ecology 171, 149e163. feeding ecology of Neoepisesarma versicolor. Estuarine, Coastal and Shelf Science Micheli, F., 1993b. Feeding ecology of mangrove crabs in North Eastern Australia: 80, 573e580. mangrove litter consumption by Sesarma messa and Sesarma smithii. Journal of Weis, J.S., Weis, P., 2004. Behavior of four species of fiddler crabs, genus Uca,in Experimental Marine Biology and Ecology 171, 165e186. southeast Sulawesi, Indonesia. Hydrobiologia 523, 47e58. More, J.W., Semmens, B.X., 2008. Incorporating uncertainty and prior information Werry, J., Lee, S.Y., 2005. Grapsid crabs mediate link between mangrove litter into stable isotope mixing models. Ecology Letters 11, 470e480. production and estuarine planktonic food chains. Marine Ecology Progress Olafsson, E., Buchmayer, S., Skov, M.W., 2002. The East African decapod crab Neo- Series 293, 165e176. sarmatium meinerti (de Man) sweeps mangrove floors clean of leaf litter. Ambio Wolcott, D.L., O’Connor, N.J., 1992. Herbivory in crabs: adaptations and ecological 31, 569e573. considerations. American Zoologist 32, 370e381.